Convolutional Neural Network based Skin Lesion Classification and Identification

Melanoma disease is the type of skin cancer in melanocytes, which are in the epidermis layer of the skin. The rate of msselanoma is in increasing order and found to be dangerous if not diagnosed at its beginning stage. To overcome this challenge, deep convolutional neural network techniques are used. The patient injured image is processing under different steps such as pre-processing using various filters followed by segmentation using the K-means clustering algorithm and Fuzzy C-means clustering algorithm. Finally, the detection and classification are executed with Convolutional Neural Network. Performance measures are evaluated for the proposed methodology with the accuracy of 98.43%, specificity of 98.77%, the sensitivity of 99.83%. The results drive that the proposed model of learning surpasses the existing algorithm and could be used to help medical practitioners to classify skin lesions.